捷联惯导系统初始对准中利用卡尔曼滤波器进行状态估计时,其数学平台的东向和北向水平失准角估计速度很快,而天向方位失准角的收敛很慢。为了解决这个问题,在基本卡尔曼滤波的基础上引入了多级组合滤波的思想,利用对捷联矩阵的修正和对天向方位失准角φU估计的改进产生了新的信息,形成了两级卡尔曼滤波,并且与小波软阈值法相结合,实现了捷联惯导系统的快速初始对准。 SINS initial alignment using Kalman filter state estimation, the mathematical platform of the east and north horizontal misalignment angle estimation speed is very fast, while the azimuth azimuth misalignment slow convergence. In order to solve this problem, based on the basic Kalman filter, the idea of ​​multi-stage combined filtering is introduced. New information is obtained by using the correction of the strapdown matrix and the improvement of the azimuth misalignment φU estimation, forming two Level Kalman filter, and combined with the wavelet soft thresholding method to achieve the rapid initial alignment of SINS.